JPH0715692B2 - Context processor - Google Patents

Description

The present invention relates to a context processing device capable of effectively determining the context structure of text in natural language processing.

(Prior Art) Recently, various studies on natural language processing have been advanced, and even the investigation of the context structure of texts and their semantic analysis have been performed. Conventionally, the determination of this kind of context structure is done by examining the nouns and verbs that appear in the text,
It is performed while referring to a dictionary in which information such as dependency conditions and relations of part-of-speech relations regarding these words is registered and checking the relations of connection of the nouns and actions. That is, the nouns and verbs appearing in the text are used as clues to obtain the context structure.

However, it is practically impossible to register information about all nouns and verbs that may appear in the text in the dictionary. It is undeniable that determining its contextual structure will be very difficult. For this reason, for example, it is necessary to create information about nouns and verbs used in the target field each time according to the content referred to by the text, and register this in the word dictionary. However, the work itself is extremely difficult, and it is difficult to adapt it to the texts of various target fields and use it as it is for determining the context structure.

In addition, text often has a rhetorical structure such as a premise and a conclusion, or a judgment and an illustration among a plurality of paragraphs. However, conventionally, as mentioned above, we are only investigating the context structure using the nouns and verbs in the text as clues.
The required context structure did not match the rhetorical structure described above, and was sometimes inappropriate.

(Problems to be Solved by the Invention) In such a conventional technique, the context structure is examined only by using the nouns and verbs that appear in the text as clues, so that the context structure is determined in the text. Some problems remain, such as having to register information about nouns and verbs in a dictionary in advance.

The present invention has been made in consideration of such circumstances, and an object thereof is to properly determine the context structure of a text even when information about individual nouns or verbs is insufficient. An object of the present invention is to provide a context processing device that is capable of achieving excellent practicality.

[Structure of the Invention] (Means for Solving the Problem) A context processing device according to the present invention comprises a correspondence table showing rhetorical expressions and connection relationships that the rhetorical expressions mean, and connection prohibition rules between the connection relationships. , A relational table used for removing connection structures that cannot be taken by a sequence of connection relations, and a sequence of connection relations between sentence parts forming the text from rhetorical expressions appearing in the text with reference to the correspondence table. A means for obtaining the rhetorical structure of the text by examining the connection structure of the series of connection relations with reference to the relation table, and a means for determining the context structure of the text by using the rhetorical structure. Is provided.

Furthermore, a relation table showing the connection regulation between the connection relations indicated by the rhetorical expression is prepared, and the connection structure of the connection relation series between the sentence parts constituting the text obtained by referring to the correspondence table is described. It is characterized in that the relational table is checked to find the rhetorical structure, and the contextual structure of the text is determined using this rhetorical structure.

(Operation) According to the present invention, even if all the information about nouns and verbs appearing in a text is not registered in the dictionary, attention is paid to the rhetorical expression in the text, and the text part is separated from the rhetorical expression. Since the rhetorical structure is determined by examining the connection structure of the connection relations and the contextual structure is determined according to this rhetorical structure, the contextual structure can be effectively determined even when information about nouns and verbs is insufficient. Is possible. Moreover, since the context structure is examined by utilizing the rhetorical structure between the sentence parts, it is possible to perform an appropriate context processing.

(Embodiment) Hereinafter, a context processing device according to an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram of a main part of the embodiment apparatus, and 1 is a text input unit for inputting text such as document data or recorded data to be analyzed. The format structure analysis unit 2 analyzes the format structure of the entire text from the format information such as chapters and sections included in the text input from the text input unit 1 and determines the structure. The morpheme analysis unit 3 executes the morpheme analysis of the sentence included in the paragraph, for example, on a paragraph-by-paragraph basis with respect to the text for which the formal structure is determined in this manner. By this morphological analysis, morphemes such as nouns and verbs in the text are individually obtained.

The above processing functions have the same functions as those of the conventional context processing device, and are realized by using various conventionally proposed methods.

Here, the feature of this device is that according to the morphological analysis result of the text obtained by the morphological analysis unit 3,
The point is that the connection relation extraction unit 4 is provided for obtaining connection relations between a plurality of sentence parts (for example, paragraphs) forming the text using the rhetorical expression in the text as a clue.
The connection relation extracting unit 4 checks the rhetorical expression in the text by referring to a correspondence table storage unit 5 which stores a rhetorical expression such as a conjunction and a correspondence table showing the connection relation as shown in FIG. It seeks the connection relation between paragraphs connected by the rhetorical expression. For example, when a rhetorical expression of "But" is requested, a connection relation indicating that the paragraph existing before the rhetorical expression and the paragraph with the above rhetorical expression have a "capture" relationship Looking for.

The connection relation extracting unit 4 sequentially finds all rhetorical expressions that appear in the text, and individually finds the connection relations between the text parts indicated by the rhetorical expressions.

Another feature of the present device in addition to the connection relation extracting unit 4 is that it has a rhetorical structure determining unit 6. The rhetorical structure determination unit 6 refers to the relation table storage unit 7 that stores, as a relation table, a prohibition rule regarding the connection of the connection relation indicated by the rhetorical expression as shown in FIG. The rhetorical structure within and between paragraphs is obtained from the series of connection relations between the sentence parts obtained as described above.

This relation table contains information indicating structural connection prohibition rules of connection relations represented by rhetorical expressions as shown in FIG. 3 (a) and negative connection relations as shown in FIG. 3 (b). It is composed of a table in which information indicating connection rules is registered, and indicates a structure between connection relationships that is inappropriate as a rhetorical structure of text. In particular, the above connection prohibition rule indicates a combinational structure that does not actually appear in the connection relation between paragraphs connected by rhetorical expressions. Also, even if the negative connection rule is not the same as the above connection prohibition rule,
Generally, it shows a combination structure that is rarely used.

The rhetorical structure determination unit 6 refers to the relation table storage unit 7 having such information, and from the sequence showing the connection relation between the paragraphs connected by the rhetorical expression obtained by the connection relation extraction unit 4, the rhetorical structure of the text. As a result, the rhetorical structure of the text is determined by finding the combination structure while eliminating inappropriate combinations. When the rhetorical structure of the text is not uniquely determined and a plurality of rhetorical structure candidates are obtained, the decision is entrusted to the next context structure determination unit 8.

Therefore, the context structure determination unit 8 basically has the knowledge storage unit 9
Using the knowledge stored in, for example, information on the field to which the content referred to by the text belongs, knowledge about the connection relation of the part of speech appearing in the text, and the dependency relation, while following the rhetorical structure of the text described above, Alternatively, determine the rhetorical structure of the text and determine its contextual structure. In other words, this context structure is basically determined in the same manner as a conventional context processing device, but at this time, the context structure is determined according to the rhetorical structure analyzed from the rhetorical expression described above. I am trying.

According to the context structure of the text obtained in this way,
The application unit 10 executes a predetermined text editing process such as creating a summary.

The above is the configuration and function of the present apparatus.

The rhetorical structure determination process based on the rhetorical expression described above will be described in more detail with reference to FIGS. 4 and 5.

FIG. 4 is a diagram showing an example of text to be subjected to context processing. When such text is input, the format structure analysis unit 2 divides the text structure constituting the text into a plurality of paragraphs or sentences from the format information such as indentation, punctuation marks, commas, periods, etc. Here, paragraphs # 1, # 2, and # 5 are respectively obtained. The morpheme analysis unit 3 executes the morpheme analysis for each paragraph thus obtained.

Then, the connection relation extracting unit 4 looks up the rhetorical expression in each paragraph by referring to the correspondence table storage unit 5 based on the morphological analysis result of each of the above paragraphs, and the connection that the rhetorical expression means for the detected rhetorical expression. Relationship information is obtained from the correspondence table storage unit 5.

In this example, as shown by the underlined wave in the figure, the rhetorical expression "one side" is obtained from paragraph # 2, and the connection relation is required to be "contrast". See also paragraph # 3
Requires the rhetorical expression "these" and its connection relation "junction". Similarly, from paragraph # 4, the display relation “for this reason” and the connection relation “junction” are obtained, and from paragraph # 5, the relation relation rhetorical expression “reference” “below” is obtained.

From such a connection relation, the connection relation extracting unit 4
As shown in FIG. 5A, the text shown in the figure requires that paragraphs # 1, # 2, and # 5 are sequentially connected in a connection relation [1 contrast 2 forward 3 forward 4 reference 5].

However, such a connection relation series merely shows connection relations among a plurality of paragraphs individually, and does not show how the rhetorical structures are combined.

Then, considering the combination of the connection relation series forming the rhetorical structure, if these five paragraphs exist, theoretically, there are 14 possible combinations as shown in FIG. 5B. When there are two paragraphs, the combination structure can take one form, and when there are three paragraphs, the combination structure can take two forms. In the case of four paragraphs, there are five types, and the number of paragraphs theoretically determines the number of forms that the combined structure can take.

The combinational structure taken by the series of connection relations indicates where each sentence portion (paragraph) in the text is connected and in what relation. For example, [1 contrast [[2 concatenation 3] order The contact [Ref 5]]] rhetorical structure (combinational structure) is the first paragraph # 1.
Indicates that there is a "contrast" relationship with the whole of the second paragraph # 2 to the fifth paragraph # 5. Then, regarding the second paragraph # 2 to the fifth paragraph # 5, the third paragraph # 3 is derived from the second paragraph # 2 in a "forward connection" relationship, and further, based on these, the fourth paragraph # 3 is introduced. It is shown that paragraph # 4 is guided in a "forward" relationship, on which the fifth paragraph # 5 follows the fourth paragraph # 4 in a "reference" relationship. The content of paragraph # 4 is shown to indicate that the relevant portion is present elsewhere in the text.

The expression of such rhetorical structure shows the connection structure relation between a plurality of paragraphs. However, the above-mentioned combination structure taken by the series of connection relations shown in B merely represents the structural form of the combination, and some of them are unsuitable as rhetorical structures in the text structure. Therefore, the rhetorical structure determination unit 6 refers to the relationship table storage unit 7 to find a structure having an inappropriate rhetorical relationship in the combined structure, and excludes this structure from candidates for rhetorical structure.

For example, as shown in FIG. 5C, the connection prohibition rule "junction [... [X order"X; Paragraph includes a combination structure of "order [3 order]" in the above combination structure. Is detected as inappropriate as a rhetorical structure. Therefore, this combination structure of the connection relation series including the combination structure of "junction [3 order"] [1 contrast [[ order [3 order [4]] 5]] is excluded from the rhetorical structure candidates. .

Similarly, the negative connection rule of the relation table includes a combination structure of connection relation series including a combination of "contrast [... [X ordering]" to "contrast [[2 ordering]" [1 contrasting [[2 ordering [3] [4 [5]]]] is excluded from the candidates for the rhetorical structure. In addition, when only this negative connection rule is satisfied, the series of the combination may be left as a rhetorical structure candidate.

In this way, the rhetorical structure determining unit 6 is inadequate as a rhetorical structure among combinations of rhetorical structures which the sequence can theoretically take in accordance with the series of the connecting relations obtained by the connecting relation extracting unit 4. Is detected / excluded by referring to the relationship table storage unit 7, and candidates as a rhetorical structure are narrowed down. Then, when the rhetorical structure is uniquely obtained in narrowing down the candidates, this is determined as the rhetorical structure, and when a plurality of candidates are left, the decision is made on the context structure determination using knowledge. Leave it to Part 8. Then, the rhetorical structure of the text determined in this way is taken into consideration, the structure of the text contents and the like are determined, and the final context structure is determined.

Thus, according to the present apparatus thus configured, not only the nouns and verbs of the text are analyzed, but also the rhetorical structure of the text indicated by the rhetorical expressions appearing in the text is analyzed, and this rhetorical structure is taken into consideration. Since the context structure of the above text is determined and context processing is performed, even if the information about the noun or verb appearing in the text is not registered in the dictionary, an appropriate context is not affected by the unregistered word. It becomes possible to perform processing. Moreover, since the context processing is performed by using the rhetorical structure analyzed from the rhetorical expression, the context processing can be performed by appropriately grasping the rhetorical structure such as premise, conclusion, judgment, and illustration. It is possible to execute the language processing that is accurately captured.

It should be noted that the present invention is not limited to the above-described embodiment, and the analysis of the format structure, the morphological analysis, etc. may be performed by appropriately using various algorithms conventionally proposed. Further, the information registered in the correspondence table or the relation table is not limited to the above example. Furthermore, although the Japanese processing has been described here, the context processing for other languages such as English can be similarly applied. In addition, the present invention can be variously modified and implemented without departing from the scope of the invention.

[Effects of the Invention] As described above, according to the present invention, the rhetorical expression that appears in a text is focused on, the rhetorical structure is obtained, and the contextual structure of the text is analyzed using this rhetorical structure. Even in the case where an unregistered word is included in the text, it is possible to effectively and accurately determine the context structure, which is very effective in practice.

[Brief description of drawings]

FIG. 1 shows a context processing device according to an embodiment of the present invention. FIG. 1 is a schematic configuration diagram of a main part of the embodiment device, FIG. 2 is a diagram showing a configuration example of a correspondence table, and FIG. FIG. 4 is a diagram showing a configuration example of a table, FIG. 4 is a diagram showing an example of text to be context-processed, and FIG. 5 is a diagram schematically showing a form of rhetorical structure analysis processing of text. 1 ... Text input part, 2 ... Format structure analysis part, 3 ...
Morphological analysis unit, 4 ... Connection relation extraction unit, 5 ... Correspondence table storage unit, 6 ... Rhetorical structure determination unit, 7 ... Relation table storage unit, 8 ... Context structure determination unit, 9 ... Knowledge storage unit ,
10 ... Application Department.

Claims (1)

[Claims] 1. A relation table used for removing a connection structure, which is composed of a correspondence table showing rhetorical expressions and connection relationships that the rhetorical expressions mean, and a connection prohibition rule between the connection relationships, which is used in a connection structure that cannot be taken by a series of connection relationships. And a means for obtaining a series of connection relations between the sentence parts constituting the text from rhetorical expressions appearing in the text by referring to the correspondence table, and a series of connection relations of the series by referring to the relation table. A context processing device comprising: means for determining a rhetorical structure of the text by examining a connection structure; and means for determining a contextual structure of the text using the rhetorical structure.